Model Predictive Control of Potassium Chloride Drying by Fluidized Bed Dryer

Potassium is an essential raw material for the development of plants, mainly as fertilizers. In a Brazilian potassium chloride plant, after tilling and flotation, the wet concentrated material follows to drying, basically comprised of a fluidized bed dryer. This drying process is non-linear, complex and influenced by several factors, thus conventional control can not be implemented properly. Advanced process control offers alternatives for that, mainly by using Model Predictive Control (MPC). The model is obtained by transient-response analysis and shows a good ability in predicting variations in temperature with changes in manipulated variables and disturbances. The linear model used by the MPC controller is periodically adjusted by an expert system with fuzzy logic accounting for the non-linear behavior of the process. A comparison is made with traditional feedback control and MPC with feed forward action.